Execute qualquer Skill no Manus
com um clique

Execute qualquer Skill no Manus com um clique

read-whatsapp

Estrelas2

Forks1

Atualizado12 de junho de 2026 às 14:20

Read local macOS WhatsApp chats and summarize user requests, including voice-note transcription. Use when the user asks Codex to inspect a WhatsApp conversation, find what someone said, understand a cancellation/request, search recent WhatsApp messages, or transcribe WhatsApp audio/voice notes from the local WhatsApp database/media folder.

Instalação

Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.

Executar no Manus

Fonte

G33KM44N38

G33KM44N38/dotfiles

Abrir repositório GitHub Ver repositórios do creator

Download

Executar no Manus

Ocupações relacionadasSOC

Baseado na classificação ocupacional SOC

Desenvolvedores de softwareInformática e Matemática·SOC 15-1252

Explorador de arquivos

3 arquivos

SKILL.md

readonly

name	read-whatsapp
description	Read local macOS WhatsApp chats and summarize user requests, including voice-note transcription. Use when the user asks Codex to inspect a WhatsApp conversation, find what someone said, understand a cancellation/request, search recent WhatsApp messages, or transcribe WhatsApp audio/voice notes from the local WhatsApp database/media folder.

Read WhatsApp

Use this skill to inspect the user's local WhatsApp Desktop data on macOS. Default to read-only database queries and local transcription.

Workflow

Locate the chat by name, phone/JID, or message text.
Pull recent messages with timestamps, sender direction, text, message type, and media paths.
When audio is present, convert it with ffmpeg and transcribe locally with whisper-cli.
Summarize only the relevant request. State transcript confidence if the model/audio is rough.

Script

Prefer the bundled script:

python3 /Users/boss/.codex/skills/read-whatsapp/scripts/read_whatsapp.py --chat "beautyhairmaiidi" --limit 80 --transcribe

Useful options:

python3 .../read_whatsapp.py --search "annuler" --limit 40
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --since "2026-06-06" --transcribe
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --transcribe --download-model

Local Transcription

Use local tools by default; no paid API.

Convert audio: ffmpeg -i input.opus -ar 16000 -ac 1 output.wav
Transcribe: whisper-cli -m ~/.cache/whisper.cpp/ggml-tiny.bin -f output.wav -l fr -otxt

If missing:

brew install whisper-cpp ffmpeg
mkdir -p ~/.cache/whisper.cpp
curl -L --fail -o ~/.cache/whisper.cpp/ggml-tiny.bin \
  https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-tiny.bin

Use larger models when exact wording matters; tiny is fast but approximate.

Data Locations

Default DB:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/ChatStorage.sqlite

Default media root:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/Message

Treat these as private user data. Do not inspect credentials. Do not modify WhatsApp files.

Mais deste repositório

mesmo repositório

explain-like-im-five

G33KM44N38/dotfiles

Create a friendly, self-contained webpage that teaches a difficult idea in very simple language using concrete analogies, small visual steps, and playful interaction. Use when the user asks for an ELI5 explanation, says "explain it like I'm five," wants a topic taught visually in a webpage, or needs a beginner-friendly interactive explainer for technical, scientific, financial, or abstract material.

2026-07-102

mailbox

G33KM44N38/dotfiles

Open and operate the user's keyboard-first Apple Mail triage desk with persistent progress and concurrent agent workers. Use when the user says mailbox, wants to review or process email one-by-one, asks to resume inbox triage, wants a local webpage for deciding what to do with messages, or wants agents to handle email instructions while they continue reviewing.

2026-07-102

natural-chat-continuation

G33KM44N38/dotfiles

Use when drafting or sending a reply in an existing chat thread, especially WhatsApp, SMS, email, or DM. Ensures the message continues the current conversation naturally by reading recent context, matching tone, and avoiding unnecessary greetings or restarts.

2026-07-092

herdr-workspace-manager

G33KM44N38/dotfiles

Manage the user's coding workspaces, threads, worktrees, tabs, panes, and running AI agents through Herdr. Use whenever the user says thread, threads, workspace, worktree, tab, pane, agent, switch, focus, open a workspace, create a worktree, start an agent, send input to an agent, check what is running/waiting, or asks to migrate old tmux thread workflows. Prefer this over tmux unless the user explicitly asks for tmux.

2026-07-062

tmux-thread-picker

G33KM44N38/dotfiles

Legacy tmux-only thread picker. Use only when the user explicitly asks for tmux or the old tmux-thread-picker CLI. For normal thread, workspace, worktree, pane, or agent management, prefer the Herdr workspace manager skill.

2026-07-062

send-whatsapp-jules

G33KM44N38/dotfiles

Draft and send WhatsApp messages to Jules from the user's macOS WhatsApp Desktop. Use when the user asks to message, notify, update, or send something to Jules on WhatsApp, especially for BabaCoiffure incident summaries; always draft the message first, apply the user's wording feedback, and send only after explicit confirmation.

2026-06-282

name	read-whatsapp
description	Read local macOS WhatsApp chats and summarize user requests, including voice-note transcription. Use when the user asks Codex to inspect a WhatsApp conversation, find what someone said, understand a cancellation/request, search recent WhatsApp messages, or transcribe WhatsApp audio/voice notes from the local WhatsApp database/media folder.

Read WhatsApp

Use this skill to inspect the user's local WhatsApp Desktop data on macOS. Default to read-only database queries and local transcription.

Workflow

Locate the chat by name, phone/JID, or message text.
Pull recent messages with timestamps, sender direction, text, message type, and media paths.
When audio is present, convert it with ffmpeg and transcribe locally with whisper-cli.
Summarize only the relevant request. State transcript confidence if the model/audio is rough.

Script

Prefer the bundled script:

python3 /Users/boss/.codex/skills/read-whatsapp/scripts/read_whatsapp.py --chat "beautyhairmaiidi" --limit 80 --transcribe

Useful options:

python3 .../read_whatsapp.py --search "annuler" --limit 40
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --since "2026-06-06" --transcribe
python3 .../read_whatsapp.py --chat "beautyhairmaiidi" --transcribe --download-model

Local Transcription

Use local tools by default; no paid API.

Convert audio: ffmpeg -i input.opus -ar 16000 -ac 1 output.wav
Transcribe: whisper-cli -m ~/.cache/whisper.cpp/ggml-tiny.bin -f output.wav -l fr -otxt

If missing:

brew install whisper-cpp ffmpeg
mkdir -p ~/.cache/whisper.cpp
curl -L --fail -o ~/.cache/whisper.cpp/ggml-tiny.bin \
  https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-tiny.bin

Use larger models when exact wording matters; tiny is fast but approximate.

Data Locations

Default DB:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/ChatStorage.sqlite

Default media root:

~/Library/Group Containers/group.net.whatsapp.WhatsApp.shared/Message

Treat these as private user data. Do not inspect credentials. Do not modify WhatsApp files.